blob: 0fb44b236e10c63b3e78113662cf192fd5d19f5d (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
|
#!/usr/bin/env python
import huggingface
import json
import os
from datasets import Dataset, Image
from glob import glob
def make_dataset(base="./baseline"): # TODO: Make actual hf dataset
prompt = "You are a professional English-German translator and also a renowned photography critic.\n\nWrite a detailed caption for this image in a single sentence. Translate the caption into German. The output needs to be JSON, the keys being 'English' and 'German' for the respective captions. Only output the JSON, nothing else." + "<start_of_image>"
user_prompts = []
images = []
assistant_replies = []
for filename in glob(f"{base}/*.jsonl"):
with open(filename, "r") as f:
data = json.loads(f.read())
image_path = f"../d/Images/{os.path.basename(filename).removesuffix(".jsonl")}.jpg"
user_prompts.append(prompt)
assistant_replies.append(json.dumps({
"English": data["English"],
"German": data["Translation"],
}, ensure_ascii=False, indent=0))
images.append(image_path)
return Dataset.from_dict({"image": images, "user": user_prompts, "assistant": assistant_replies}).cast_column("image", Image())
def main():
huggingface.login()
dataset = make_dataset()
dataset.push_to_repo("asdf2k/caption_translation")
if __name__ == "__main__":
main()
|
